JP6440543B2 - Electric power demand prediction apparatus and electric power demand prediction method - Google Patents

Description

  The present invention relates to a power demand prediction apparatus and a power demand prediction method for predicting a power demand amount of a facility that performs sewage treatment such as a sewer pump station or a water reclamation center that treats rainwater and wastewater in a combined manner. It is possible to accurately predict fluctuations in the amount of power demand due to the above.

  In recent years, the introduction of distributed power sources such as storage batteries or generators has progressed, and energy management systems for controlling them in cooperation have been developed. In order to optimally control distributed power sources in a coordinated manner, it is necessary to accurately grasp the power demand in customer facilities. So far, we have built a prediction model that models the relationship between weather information and power demand. Methods for predicting demand have been proposed.

  For example, in the power demand prediction method described in Patent Document 1, a prediction model is constructed from the relationship between the temperature and the power amount at each time, and the power demand amount is predicted using the prediction model. Moreover, in the power demand prediction method described in Patent Document 2, a prediction model is constructed using weather data such as weather and rainfall to predict power demand.

JP 2004-164388 A JP 2013-161336 A

  However, the sewage treatment facility has a role of treating rainwater as well as wastewater treatment. Normally, rainwater is discharged from a rainwater discharge pump to the sea or river via a sewer main line. For this reason, the power demand at the sewage treatment facility has little relationship with the air temperature, and the power demand prediction method disclosed in Patent Document 1 cannot be accurately predicted.

  In addition, as in Patent Document 2, the power demand can be predicted to some extent by using rainfall and precipitation. However, in sewage treatment facilities, in addition to facilities such as lighting and air conditioning provided in general facilities, Facilities for treating rain such as pumps and storage tanks are provided. The pump consumes electric power when pumping up rainwater flowing into the storage tank.

  Therefore, the power demand of the entire facility changes due to the influence of the operation status of water treatment facilities such as pumps. The operation status of water treatment facilities such as pumps changes depending on the cumulative value such as how much rain has fallen in the past few hours and how much rain will fall in the next few hours. The future operating situation will change depending on whether the water treatment facility is operated. For this reason, there is a problem in that accurate prediction cannot be made by simply predicting power demand based on the relationship between rainfall and power demand.

  The present invention has been made to solve the above-described problems, and provides a power demand prediction apparatus and a power demand prediction method that accurately predict fluctuations in power demand due to rainwater in a facility that performs sewage treatment. With the goal.

The power demand prediction apparatus of the present invention is
In the power demand forecasting device that predicts the predicted amount of power at the forecast target time in a facility that performs sewage and rainwater treatment,
The actual power demand value at each time of the facility is acquired, the maximum actual power demand value from the first predetermined time before the target time to the target time is detected with respect to the actual power demand value, and the target time An actual data acquisition unit that extracts the maximum electric energy value of
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. The cumulative precipitation of each precipitation value at each time from the target time to the third predetermined time after the target time is calculated by calculating the cumulative precipitation of the quantity value and extracting it as the past precipitation at the target time. A meteorological information acquisition unit that calculates and extracts as future precipitation at the target time;
A maximum power amount value database for storing the maximum power amount value at the target time;
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
A demand prediction unit that predicts the predicted power amount at the prediction target time from the relationship between the maximum power amount value, the past precipitation amount, and the future precipitation amount and the actual power demand value.

Moreover, the power demand prediction apparatus of this invention is
In the power demand prediction device that predicts the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
A record data acquisition unit that acquires a power demand record value at each time of the facility;
A pump data acquisition unit that extracts the operation state of the pump from the fourth predetermined time before the target time to the target time as the target operation state;
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. A cumulative amount of precipitation is calculated and extracted as past precipitation at the target time, and the cumulative precipitation amount at each time from the target time to the third predetermined time after the target time is calculated. A meteorological information acquisition unit for extracting as future precipitation at the target time;
A pump operating state database for accumulating the target operating state at the target time;
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
The apparatus includes a demand prediction unit that predicts the predicted electric energy amount at the prediction target time from the relationship between the target operation state, the past precipitation amount, and the future precipitation amount, and the actual power demand value.

Moreover, the power demand prediction apparatus of this invention is
In the power demand prediction device that predicts the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
Obtain the first power demand actual value of the pump and the second power demand actual value other than the pump of the facility,
A record of detecting the maximum first power demand record value from the fifth predetermined time before the target time to the target time with respect to the first power demand record value and extracting it as the pump maximum power amount value at the target time A data acquisition unit;
Obtaining weather information including precipitation values and temperature values of actual values and forecast values related to the facility, extracting a temperature value of the target time for the weather information, and a first value of the target time (2) calculating a cumulative precipitation of each precipitation value at each time from a predetermined time before to the target time and extracting it as a past precipitation at the target time; and a third predetermined time from the target time to the target time A meteorological information acquisition unit that calculates a cumulative precipitation of each precipitation value at each time until a time later and extracts it as a future precipitation at the target time;
A pump maximum electric energy value database for storing the pump maximum electric energy value at the target time;
A temperature value database for accumulating temperature values at the target time; and
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
Predicting the first electric power prediction value of the pump at the prediction target time from the relationship between the pump maximum electric energy value, the past precipitation, and the future precipitation, and the first electric power demand actual value, and the temperature value And predicting the second power amount predicted value other than the pump at the prediction target time from the relationship between the second power demand actual value and the total of the first power amount predicted value and the second power amount predicted value. A demand prediction unit that predicts the value as the electric energy prediction value.

The power demand prediction method of the present invention is
In the power demand prediction method for predicting the predicted amount of electric power at the prediction target time in a facility that performs sewage and rainwater treatment,
The actual power demand value at each time of the facility is acquired, the maximum actual power demand value from the first predetermined time before the target time to the target time is detected with respect to the actual power demand value, and the target time Extracted as the maximum power value of
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. The cumulative precipitation of each precipitation value at each time from the target time to the third predetermined time after the target time is calculated by calculating the cumulative precipitation of the quantity value and extracting it as the past precipitation at the target time. Is calculated and extracted as future precipitation at the target time,
The power amount prediction value at the prediction target time is predicted from the relationship between the maximum power amount value, the past precipitation amount, and the future precipitation amount and the actual power demand actual value.

The power demand prediction method of the present invention is
In the power demand prediction method for predicting the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
A record data acquisition unit that acquires a power demand record value at each time of the facility;
The operation state of the pump from the fourth predetermined time before the target time to the target time is extracted as the target operation state,
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. A cumulative amount of precipitation is calculated and extracted as past precipitation at the target time, and the cumulative precipitation amount at each time from the target time to the third predetermined time after the target time is calculated. And extract it as future precipitation at the target time,
The predicted power amount at the prediction target time is predicted from the relationship between the target operating state, the past precipitation amount, and the future precipitation amount and the actual power demand value.

The power demand prediction method of the present invention is
In the power demand prediction method for predicting the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
The first power demand record value of the pump and the second power demand record value other than the pump of the facility are acquired, and the target from the fifth predetermined time before the target time with respect to the first power demand record value The maximum first power demand actual value until the time is detected and extracted as the pump maximum power amount value at the target time,
Obtaining weather information including precipitation values and temperature values of actual values and forecast values related to the facility, extracting a temperature value of the target time for the weather information, and a first value of the target time (2) calculating a cumulative precipitation of each precipitation value at each time from a predetermined time before to the target time and extracting it as a past precipitation at the target time; and a third predetermined time from the target time to the target time Calculate the cumulative precipitation of each precipitation value at each time until time later and extract it as future precipitation at the target time,
Predicting the first electric power prediction value of the pump at the prediction target time from the relationship between the pump maximum electric energy value, the past precipitation, and the future precipitation, and the first electric power demand actual value, and the temperature value And predicting the second power amount predicted value other than the pump at the prediction target time from the relationship between the second power demand actual value and the total of the first power amount predicted value and the second power amount predicted value. Value The value is predicted as the power amount prediction value.

  According to the power demand prediction device and the power demand prediction method of the present invention, it is possible to accurately predict fluctuations in the power demand due to rainwater.

It is a figure which shows the structure of the electric power demand prediction apparatus of Embodiment 1 of this invention. It is a figure for demonstrating the maximum electric energy value in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the past precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the future precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the prediction model of the neural network in the demand prediction part of the electric power demand prediction apparatus shown in FIG. It is a flowchart for demonstrating operation | movement of the demand prediction part of the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the maximum electric energy value in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the maximum electric energy value in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the past precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the past precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the future precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the future precipitation in the electric power demand prediction apparatus shown in FIG. It is a figure which shows the structure of the electric power demand prediction apparatus of Embodiment 2 of this invention. It is a figure for demonstrating the target driving | running state of the pump in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the prediction model of the neural network in the demand prediction part of the power demand prediction apparatus shown in FIG. It is a flowchart for demonstrating operation | movement of the demand prediction part of the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the target driving | running state of the pump in the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the target driving | running state of the pump in the electric power demand prediction apparatus shown in FIG. It is a figure which shows the structure of the electric power demand prediction apparatus of Embodiment 3 of this invention. It is a figure for demonstrating the prediction model of the neural network in the demand prediction part of the electric power demand prediction apparatus shown in FIG. It is a figure for demonstrating the prediction model of the neural network in the demand prediction part of the electric power demand prediction apparatus shown in FIG. It is a flowchart for demonstrating operation | movement of the demand prediction part of the electric power demand prediction apparatus shown in FIG.

Embodiment 1 FIG.
Embodiments of the present invention will be described below. 1 is a block diagram showing a configuration of a power demand prediction apparatus according to Embodiment 1 of the present invention.
FIG. 2 is a diagram for explaining the maximum power amount value at the target time in the power demand prediction apparatus shown in FIG.
FIG. 3 is a diagram for explaining the past precipitation amount at the target time in the power demand prediction apparatus shown in FIG.
FIG. 4 is a diagram for explaining the future precipitation amount at the target time in the power demand prediction apparatus shown in FIG.

FIG. 5 is a diagram for explaining a neural network prediction model in the demand prediction unit of the power demand prediction apparatus shown in FIG.
FIG. 6 is a flowchart for explaining the operation of the demand prediction unit of the power demand prediction apparatus shown in FIG.
FIG. 7 is a diagram for explaining the maximum power amount value at the prediction start time in the power demand prediction apparatus shown in FIG.
FIG. 8 is a diagram for explaining the maximum power amount value at the prediction target time in the power demand prediction apparatus shown in FIG.

FIG. 9 is a diagram for explaining the past precipitation amount at the prediction start time in the power demand prediction apparatus shown in FIG.
FIG. 10 is a diagram for explaining the past precipitation amount at the prediction target time in the power demand prediction apparatus shown in FIG.
FIG. 11 is a diagram for explaining the future precipitation amount at the prediction start time in the power demand prediction apparatus shown in FIG.
FIG. 12 is a diagram for explaining the future precipitation amount at the prediction target time in the power demand prediction apparatus shown in FIG.

  In FIG. 1, a power demand prediction apparatus 1 predicts a power demand amount of a facility 2 that performs sewage treatment such as a sewer pumping station or a water reclamation center that treats rainwater and wastewater in a combined manner. In the facility 2, for example, there are facilities that use electric power, such as a pump, a blower, lighting, and air conditioning. Therefore, the power demand amount of the facility 2 varies depending on the usage amount of these facilities.

  The power demand prediction apparatus 1 includes a performance data acquisition unit 50, a weather information acquisition unit 30, a demand prediction unit 40, a power demand performance value database (hereinafter abbreviated as DB) 11, and a maximum power amount value. DB12, precipitation DB13, past precipitation DB14, and future precipitation DB15 are provided.

  The actual data acquisition unit 50 acquires the actual power value used in the facility 2 as the actual power demand value EN, and stores the actual power demand value EN for each time in the actual power demand value DB 11. Note that here, every hour, data is acquired for each time. Further, the actual data acquisition unit 50 determines the maximum power demand actual value EN at each time from the power demand actual value EN in the power demand actual value DB11 to the target time TN from the first predetermined time M1 before the target time TN. Is detected as the maximum power amount value GN at the target time TN. And the performance data acquisition part 50 stores in the largest electric energy value DB12 as the largest electric energy value GN of the object time TN.

  The meteorological information acquisition unit 30 acquires meteorological information 3 including the actual precipitation value and the predicted precipitation value from the outside, and stores the precipitation value SN for each time in the precipitation DB 13. The weather information 3 related to the facility 2 may be acquired from a company that provides external weather information or may be acquired from a sensor installed in the facility 2. The weather information 3 related to the facility 2 refers to the weather information 3 in a range considered to flow into the facility 2 as rainwater. The range is set as appropriate depending on the scale of the facility 2 and the results of prediction. In addition, every time here is every time that is the same as the previous power demand record value EN, and data for every hour is acquired.

  Further, the meteorological information acquisition unit 30 accumulates the precipitation amount SN of each precipitation value SN at each time from the precipitation value SN at each time of the precipitation DB 13 to the target time TN from the second predetermined time M2 before the target time TN. Is extracted as the past precipitation amount RN at the target time TN. And the weather information acquisition part 30 stores in the past precipitation DB14 as the past precipitation RN of the object time TN.

  Furthermore, the meteorological information acquisition unit 30 accumulates the precipitation amount SN of each precipitation value SN at each time from the precipitation value SN at each time in the precipitation DB 13 to the third predetermined time M3 after the target time TN to the target time TN. And is extracted as the future precipitation QN at the target time TN. And the weather information acquisition part 30 stores in the future precipitation DB15 as the future precipitation QN of the target time TN.

  The demand prediction unit 40 includes a power demand actual value DB11, a maximum power amount value DB12, a past precipitation amount DB14, and a maximum power amount value GN, a past precipitation amount RN at a target time TN accumulated in the future precipitation amount DB15, and The predicted electric energy value FX at the prediction target time TX is predicted from the relationship between the future precipitation QN and the actual power demand value EN.

  A method for predicting the predicted power amount FX at the prediction target time TX of the power demand prediction apparatus 1 of the first embodiment configured as described above will be described. First, before performing the prediction, for example, using the data of the facility 2 and the weather information 3 over the past several years, the maximum electric energy value GN, the past precipitation RN, and the future precipitation QN at each target time TN. Is extracted or calculated to obtain the relationship. If there is data for several years, it is appropriate because the relationship of seasonal changes can be obtained, but not limited to this, it is calculated using data for several days or months. However, it is possible to make the prediction. In addition, since this is the same also in the following embodiment, the description is abbreviate | omitted suitably.

  Next, the performance data acquisition unit 50 acquires the actual power value used in the facility 2 as the power demand actual value EN every hour, and stores the actual power demand actual value EN for each time in the power demand actual value DB 11. Store. Then, as shown in FIG. 2, the performance data acquisition unit 50, for example, 4 hours (4H) before the first predetermined time M1 of the target time TN, from the power demand performance value EN of the power demand performance value DB 11 as shown in FIG. The power demand record value EN at each time from the target time TN-M1 to the target time TN is extracted. Then, the maximum power demand actual value EN in each power demand actual value EN is detected and extracted as the maximum power amount value GN at the target time TN.

  And the performance data acquisition part 50 stores in the largest electric energy value DB12 as the largest electric energy value GN of the object time TN. In this way, with each time as the target time TN, the maximum electric energy value GN at each time (target time TN) is detected and stored in association with each target time TN.

Next, the weather information acquisition unit 30 acquires the weather information 3 including the actual precipitation value for each hour and the forecast value for the precipitation from the outside, and stores the precipitation value SN for each hour in the precipitation DB 13. To do.
Then, as shown in FIG. 3, the meteorological information acquisition unit 30 determines, for example, 7 hours (7H) before the second predetermined time M2 of the target time TN, as shown in FIG. Cumulative precipitation of each precipitation value SN at each time from the target time TN-M2 to the target time TN (however, not including the target time TN) is calculated. And it extracts as the past precipitation RN of object time TN.

  Here, when calculating the precipitation amount RN in the past, when the precipitation value SN for each time is not weighted, or when multiplying the weight value αL to weight the precipitation value SN for each time Can be considered.

When weighting is not performed at each time, since the second predetermined time M2 is 7 hours here, the weight αL is
{Α0, α1, α2, α3, α4, α5, α6} = {1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0}
It is conceivable that all take the same value of 1.0 and do not weight each time.
The subscript L of the weight αL takes values of 0, 1, 2, 3, 4, 5, 6 from the time T1 one hour before the target time TN to the target time TN-M2.

In addition, when the precipitation value SN at each time is weighted, the weight αL takes a real value from 0.0 to 1.0,
{Α0, α1, α2, α3, α4, α5, α6} = {0.0, 1.0, 0.75, 0.5, 0.25, 0.125, 0.0625}
In this way, it is conceivable to take a weight αL of a different weight value depending on each time. Therefore, the past precipitation amount RN can be obtained by (Equation 1) shown below.

  This weighting value is appropriately changed and set according to the relationship between each facility 2 and the amount of precipitation, and generally, the weight αL is considered to increase with respect to a time close to the target time TN. It is done. However, in this case, it is assumed that the precipitation at the past time T1 one hour before the target time TN has not reached the facility 2 and is applied to the past precipitation RN with a weight α0 = 0. The idea of not accumulating automatically is shown.

  And the weather information acquisition part 30 stores in the past precipitation DB14 as the past precipitation RN of the object time TN. In this way, the past precipitation amount RN at each time (target time TN) is calculated with each time as the target time TN, and stored in association with each target time TN.

  Further, as shown in FIG. 4, the meteorological information acquisition unit 30 determines, for example, 6 hours before the third predetermined time M3 from the target time TN to the target time TN, as shown in FIG. (6H) The cumulative precipitation amount of each precipitation value SN at each time up to the target time TN + M3 after is calculated. And it extracts as future precipitation amount QN of object time TN.

  Here, when calculating the precipitation QN in the future, when the precipitation value SN for each time is not weighted, or when multiplying the weight βK to weight the precipitation value SN for each time Can be considered.

When weighting is not performed at each time, since the third predetermined time M3 is 6 hours here, the weight βK is
{Β0, β1, β2, β3, β4, β5} = {1.0, 1.0, 1.0, 1.0, 1.0, 1.0}
It is conceivable that all take the same value as 1.0 and do not weight each time.
The subscript K of the weight βK takes values 0, 1, 2, 3, 4, 5 from the target time TN to the target time TN + M3.

In addition, when weighting the precipitation value SN at each time, the weight βK takes a real value from 0.0 to 1.0,
{Β0, β1, β2, β3, β4, β5} = {1.0, 0.75, 0.5, 0.25, 0.125, 0.0625}
In this way, it is conceivable to take a weight βK having a different weight value depending on each time. Therefore, the future precipitation QN can be obtained by (Equation 2) shown below.

  This weighting value is appropriately changed and set according to the relationship between each facility 2 and the amount of precipitation. In general, the weight βK is considered to increase with respect to a time close to the target time TN. It is done.

  And the weather information acquisition part 30 stores in the future precipitation DB15 as the future precipitation QN of the target time TN. In this way, with each time as the target time TN, the future precipitation QN at each time (target time TN) is calculated and stored in association with the target time TN.

  In the example shown here, the precipitation value SN at the target time TN is not accumulated in the past precipitation RN, but is accumulated in the future precipitation QN. However, the present invention is not limited to this. A method is also conceivable in which the past precipitation RN is added but not the future precipitation QN.

Next, the demand prediction unit 40 will be described.
Using each value of the past precipitation DB 14, future precipitation DB 15, maximum power amount value DB 12, and actual power demand value DB 11 for each past target time TN determined as described above, for each target time TN. Build a predictive model. And the electric energy prediction value FX of the prediction object time TX is calculated from the past precipitation RX of the prediction object time TX, the future precipitation QX, and the maximum electric energy value GX.

  In addition, as a calculation method of the electric energy prediction value FX, the method of calculating using the prediction model by a multiple regression analysis or a neural network can be considered. A method using a pattern matching method is also conceivable.

First, as a specific example, a method for calculating an electric energy prediction value using a prediction model based on multiple regression analysis will be described.
A formula for calculating the predicted electric energy FX at the prediction target time TX using the prediction model is shown below as (Formula 3).

  Predicted electric energy (FX) = PP · Past precipitation (RN) + PF · Future precipitation (QN) + MAX · Maximum electric energy (GN) + intercept (Equation 3)

  Here, PP, PF, and MAX are partial regression coefficients, and intercept represents an intercept. The constant regression coefficient and the intercept are obtained by using the least square method so that a constant term that minimizes the sum of squares of errors in (Equation 4) shown below is obtained.

Next, as a specific example, a method for calculating a predicted electric energy value using a prediction model based on a neural network will be described.
An example of a neural network prediction model is shown in FIG.
As illustrated in FIG. 5, the predicted power amount FX at the prediction target time TX is calculated using the prediction model with the past precipitation RX at the prediction target time TX, the future precipitation QX, and the maximum power amount GX as input values. . The formula for calculating the electric energy predicted value FX is shown below as (Formula 5).

  Here, Xj represents an input value, ie, past precipitation RX, future precipitation QX, and maximum power value GX, wi is a coupling coefficient of each unit connecting the output layer to the intermediate layer, and wij is an input layer and an intermediate layer , F represents an output function in the output layer, and fj represents an output function in the intermediate layer. The coupling coefficients wi and wij of each unit are calculated by the least square method from the relationship between the past precipitation RN, the future precipitation QN, the maximum power amount value GN, and the actual power demand value EN at each past time. .

  Next, as a specific example, a method for calculating the predicted electric energy FX using a pattern matching method will be described. FIG. 6 is a flowchart showing a processing procedure of power demand prediction in the demand prediction unit 40.

First, a narrowing-down condition for acquiring the power demand actual value EN used for power demand prediction from the past power demand actual value DB 11 is acquired (step ST1 in FIG. 6).
For example, as the narrowing-down conditions, the following examples using the past precipitation amount RN, the future precipitation amount QN, and the maximum power amount value GN can be considered.

| Past precipitation RX at the prediction target time TX−Past precipitation RN at the target time TN to be compared | ≦ A1 = 1 mm
| Future precipitation QX at forecast time TX-Future precipitation QN at time TN to be compared | ≦ A2 = 2 mm
| Maximum electric energy value GX at prediction target time TX−Maximum electric energy value GN at target time TN to be compared | ≦ A3 = 500 kW
As described above, each expression is used as a narrowing condition.

  The prediction target time TX is an arbitrary time at which future power demand including the current time is predicted. The comparison target time TN is an arbitrary past time that does not include the current time. In addition, A1, A2, and A3 are set to 1 mm, 5 mm, and 500 kW, respectively. However, it is not limited to this, and takes a value of 0 or more depending on the facility configuration of the facility 2 and the possible amount of wastewater treatment. Because of differences, accuracy evaluation is performed using past data when the system is introduced, and the best value is set.

Next, the actual power demand value EN at the comparison target time TN corresponding to the narrowing condition is acquired from the actual power demand value DB 11 based on the narrowing condition (step ST2 in FIG. 6).
Next, an average value of the acquired power demand actual value EN is calculated, and this value is set as the predicted power amount FX at the prediction target time TX (step ST3 in FIG. 6).

  As described above, the power amount predicted value FX at the prediction target time TX predicted by the demand prediction unit 40 using each method is stored in the power demand actual value DB 11 as the power demand actual value EX at the prediction target time TX. Store. Note that the power demand actual value EX stored based on the prediction of the prediction target time TX is obtained when the actual time reaches the prediction target time TX and the actual power demand actual value EX can be obtained. In this case, the actual power demand actual value EX is replaced and stored.

  Hereinafter, a case where prediction is actually performed will be further described. As shown in FIG. 7, the actual power demand value EN up to the prediction target time TX, which is the time at which prediction is to be performed, at the prediction start time TY, which is the time at which prediction is performed, includes the prediction start time TY, including the prediction start time TY. Does not exist until.

  Therefore, first, the electric energy prediction value FY at the prediction start time TY is calculated by the method described above. And as shown in FIG. 8, this electric energy predicted value FY is registered into electric power demand actual value DB11 as the electric energy predicted value FY of prediction start time TY. Then, the predicted electric energy value FX corresponding to each target time TN from the prediction start time TY to the prediction target time TX is sequentially obtained.

  In this case, the past precipitation RX at the prediction target time TX will be described. As shown in FIG. 9, at the prediction start time TY, the past precipitation amount RY is obtained only from the precipitation value SN as the actual value. However, as shown in FIG. 10, the precipitation value SN at each time from the second predetermined time M2 hours before the prediction target time TX to the prediction target time TX has no actual value, and is calculated using the prediction value. It will be.

  Further, the future precipitation QX at the prediction target time TX will be described. As shown in FIG. 11 and FIG. 12, since there is no actual precipitation value at any of the prediction start time TY and the prediction target time TX, the predicted value of the precipitation value SN at each time is used to determine the future. It is calculated as precipitation QY and future precipitation QX.

According to the power demand prediction apparatus of Embodiment 1 configured as described above,
The relationship between the cumulative rainfall from the predetermined time before the forecast target time, the cumulative precipitation from the forecast target time to the predetermined time ahead, the maximum power value from the predetermined time to the forecast target time, and the actual power demand value Since the prediction of the predicted power amount at the prediction target time is performed, it is possible to accurately predict the power demand taking into account the fluctuation of rainwater in the facility.

  In addition, the demand prediction unit stores the power amount prediction value at the prediction target time in the power demand actual value DB as the power demand actual value at the prediction target time, and the actual data acquisition unit sets as the power demand actual value at the prediction target time. Therefore, the electric energy prediction value at the prediction target time can be sequentially performed.

  In addition, since the past precipitation and future precipitation are calculated by weighting the precipitation values at each target time, it is possible to more accurately predict the electricity demand taking into account the fluctuation of rainwater according to the facility situation. Can do.

  In addition, each parameter (value, such as predetermined time, every time, a weight) shown in the said Embodiment 1 changes with facilities installation structures, the possible amount of waste water treatment, etc. For this reason, each parameter is subjected to accuracy evaluation using past data, and a value considered to be appropriate is set.

Further, each parameter is not limited to the one set at the beginning, but may be appropriately changed according to the evaluation of the prediction of the electric energy.
This also applies to the following embodiments, and the description thereof will be omitted as appropriate.

Embodiment 2. FIG.
FIG. 13 is a block diagram showing the configuration of the power demand prediction apparatus according to Embodiment 2 of the present invention.
FIG. 14 is a diagram for explaining the target operation state of the pump at the target time of the power demand prediction apparatus shown in FIG.
FIG. 15 is a diagram for explaining a neural network prediction model in the demand prediction unit of the power demand prediction apparatus shown in FIG.

FIG. 16 is a flowchart for explaining the operation of the demand prediction unit of the power demand prediction apparatus shown in FIG.
FIG. 17 is a diagram for explaining the target operation state of the pump at the prediction start time of the power demand prediction apparatus shown in FIG.
FIG. 18 is a diagram for explaining the target operation state of the pump at the prediction target time of the power demand prediction apparatus shown in FIG.
The same parts as those in the first embodiment are denoted by the same reference numerals and the description thereof is omitted.

In FIG. 13, the power demand prediction apparatus 1 includes a pump data acquisition unit 60 and a pump operation state DB 16. The pump data acquisition unit 60 acquires the operation state of the pump of the facility 2 and stores the operation state of the pump for each time in the pump operation state DB 16. Here, as for each time, it is assumed that data for every hour is acquired as in the first embodiment.
Further, the pump data acquisition unit 60 extracts the operation state of the pump from the operation state of the pump of the pump operation state DB 16 to the target time TN from the fourth predetermined time M4 before the target time TN as the target operation state of the target time TN. And stored in the pump operation state DB 16.

Next, operation | movement of the electric power demand prediction apparatus 1 of Embodiment 2 comprised as mentioned above is demonstrated. Note that the same operations as those in the first embodiment are omitted as appropriate.
First, the pump data acquisition unit 60 acquires, as shown in FIG. 14, for example, operation information on the number of pumps that are turned on or off for each hour as the pump operation state of the facility 2.

  And the pump data acquisition part 60 is the operation state of the pump of each time from the target time TN-7H before the fourth predetermined time M4 of the target time TN, for example, 7 hours (7H) before the target time TN, here. Extract. In this case, the maximum activation number VN is detected as the target operation state among the operation states of each pump, and is extracted as the maximum activation number VN at the target time TN.

  Then, the pump data acquisition unit 60 stores the target operation state at the target time TN as the maximum activation number VN in the pump operation state DB 16. In this way, with each time as the target time TN, the maximum activation number VN at each time (target time TN) is detected and stored in association with each target time TN.

  In addition, although the case where the maximum starting number VN of the pump starting numbers at each time from the fourth predetermined time M4 hours before the target time TN to the target time TN is set as the target operation state at the target time TN is shown. However, the present invention is not limited to this. For example, the total number of pump activations at each time from M4 hours before the target time TN to the target time TN may be set.

  Next, the demand prediction unit 40 will be described. Since the past precipitation RN and the future precipitation QN at the target time TN are calculated in the same manner as in the first embodiment, the description thereof is omitted. A prediction model for each target time TN using the values of the past precipitation DB 14, future precipitation DB 15, pump operation state DB 16, and actual power demand value DB 11 for each past target time TN obtained as described above. Build up. Then, the electric power prediction value FX at the prediction target time TX is calculated from the past precipitation RX at the prediction target time TX, the future precipitation QX, and the maximum activation number VX.

  As a method for calculating the electric energy predicted value FX, a method of calculating using a regression model or a prediction model by a neural network is conceivable as in the first embodiment. A method using a pattern matching method is also conceivable.

  First, as a specific example, a method for calculating an electric energy prediction value using a prediction model based on multiple regression analysis will be described. A formula for calculating the predicted electric energy FX at the prediction target time TX using the prediction model is shown below as (Formula 6).

Predicted electric energy (FX) = PP, past precipitation (RN) + PF, future precipitation (QN) + pump, pump activation state (VN) + intercept (Equation 6)

  Here, PP, PF, and pomp are partial regression coefficients, and intercept represents an intercept. The constant regression coefficient and the intercept are obtained by using the least square method so that a constant term that minimizes the sum of squares of errors in (Equation 7) shown below is obtained.

  Next, as a specific example, a method for calculating a predicted electric energy value using a prediction model based on a neural network will be described. An example of a neural network prediction model is shown in FIG. As illustrated in FIG. 15, the power amount prediction value FX at the prediction target time TX is calculated using the prediction model with the past precipitation RX at the prediction target time TX, the future precipitation QX, and the maximum activation number VX as input values. A formula for calculating the electric energy prediction value FX is shown below as (Formula 8).

  Here, Xj represents the input value, the past precipitation RX, the future precipitation QX, and the maximum activation number VX, wi is a coupling coefficient of each unit connecting the intermediate layer to the output layer, and wij is the input layer and the intermediate layer. The coupling coefficient of each unit to be connected, f is an output function in the output layer, and fj is an output function in the intermediate layer. The coupling coefficients wi and wij of each unit are calculated by the least square method from the relationship between the past precipitation RN, the future precipitation QN, the maximum startup number VN, and the actual power demand value EN at each past time.

  Next, as a specific example, a method for calculating a predicted electric energy using a pattern matching method will be described. FIG. 16 is a flowchart showing a processing procedure of power demand prediction in the demand prediction unit 40.

First, a narrowing-down condition for acquiring the power demand actual value EN used for power demand prediction from the past power demand actual value DB 11 is acquired (step ST11 in FIG. 16).
For example, as the narrowing-down conditions, the following examples using the past precipitation RN, the future precipitation QN, and the maximum activation number VN can be considered.

| Past precipitation RX at the prediction target time TX−Past precipitation RN at the target time TN to be compared | ≦ A4 = 1 mm
| Future precipitation QX at forecast time TX-Future precipitation QN at time TN to be compared | ≦ A2 = 5 mm
Maximum number of activations VX at prediction target time TX = Maximum number of activations VN at target time TN to be compared
As described above, each expression is used as a narrowing condition.

  The prediction target time TX is an arbitrary time at which future power demand including the current time is predicted. The target time TN to be compared is an arbitrary past time that does not include the current time. In addition, A4 and A5 are set to 1 mm and 5 mm, respectively. However, the present invention is not limited to this, and takes a value of 0 or more and varies depending on the equipment configuration of the facility 2 and the amount of wastewater treatment. At the time of introduction, the past data will be used to evaluate accuracy and set the best value.

Next, the actual power demand value EN at the comparison target time TN corresponding to the narrowing condition is acquired from the actual power demand value DB 11 based on the narrowing condition (step ST12 in FIG. 16).
Next, an average value of the acquired power demand actual value EN is calculated, and this value is set as the predicted power amount FX at the prediction target time TX (step ST13 in FIG. 16).

  As described above, the power amount predicted value FX at the prediction target time TX predicted by the demand prediction unit 40 using each method is stored in the power demand actual value DB 11 as the power demand actual value EX at the prediction target time TX. Store. Note that the power demand actual value EX stored based on the prediction of the prediction target time TX is obtained when the actual time reaches the prediction target time TX and the actual power demand actual value EX can be obtained. In this case, the actual power demand actual value EX is replaced and stored.

  Hereinafter, a case where prediction is actually performed will be further described. As shown in FIG. 17, the maximum number of activations VN from the prediction start time TY to the prediction target time TX at the prediction start time TY that is the time to perform prediction is the pump from the prediction start time TY to the prediction target time TX. Since there is no information on the driving state of, it cannot be detected. Note that the maximum activation number VY at the predicted start time TY can be extracted as “2”. Therefore, in order to simply predict the maximum number of activations VX until the prediction target time TX thereafter, as shown in FIG. 18, the maximum number of activations VN to VX is calculated from the maximum number of activations VN to the prediction target time TX including the prediction start time TY. It is also conceivable to set each to 0.

  In addition, as a method for obtaining the maximum activation number VN from the prediction start time TY to the prediction target time TX, first, the predicted electric energy value FY at the prediction start time TY is calculated by the method described above. And this electric energy prediction value FY is registered into power demand track record value DB11 as power demand track record value EY of prediction start time TY.

  Then, the pump data acquisition unit 60 predicts the operation state of the pump at the prediction start time TY, that is, the number of pumps operated, from the predicted electric energy amount FY (= the actual power demand value EY) at the prediction start time TY. In this case, for example, the prediction can be made based on the following conditional expression.

0 pumps: 0 ≦ predicted electric energy FY ≦ threshold γ1
One pump: threshold γ1 ≦ predicted electric energy FY ≦ threshold γ2
Two pumps: threshold γ2 ≦ predicted electric energy FY ≦ threshold γ3
...
Pump n units: threshold value γn ≦ predicted electric energy FY
Note that the threshold value γn exists for the number of pumps of the facility 2, and a value estimated from the past actual power demand actual value EN and the number of operating pumps is set as the threshold value.

Specifically, a case will be described in which three pumps, a first pump, a second pump, and a third pump exist in the facility 2 and the electric energy at the time of starting each pump is shown below.
When starting the first pump = 1000kW
When starting the second pump = 1500 kW
When the third pump is started = 800 kW

Therefore, the conditional expressions shown above are set as shown below.
0 pumps: 0 ≦ predicted electric energy FY <800 kW
One pump: 800 kW ≦ power predicted value FY <1800 kW
Two pumps: 1800 kW ≦ predicted electric energy FY <2500 kW
3 pumps: 2500 kW ≤ predicted electric energy FY

  Therefore, for example, if the electric energy prediction value FY is 1000 kW, the number of pumps started is estimated to be one. Then, the calculated operation state of the pump at the predicted start time TY is stored in the pump operation state DB 16. In this manner, the pump operating state corresponding to each target time TN up to the prediction target time TX is sequentially obtained, and based on this, the predicted electric energy FX corresponding to each target time TN up to the prediction target time TX is calculated. Are obtained sequentially.

  According to the power demand prediction apparatus of the second embodiment configured as described above, the accumulated precipitation from the predetermined time before the prediction target time, as well as the same effect as the first embodiment, Using the relationship between the accumulated precipitation from the prediction target time to the predetermined time ahead, the target operation state of the pump from the predetermined time before the prediction target time, and the actual power demand value, Since the prediction is performed, it is possible to accurately predict the power demand in consideration of the rainwater fluctuation of the facility.

  Moreover, since the pump data acquisition unit can extract and set the target operation state as the maximum number of pump activations or the total number of pump activations, the target operation state can be easily set.

  Furthermore, since the operation state of the pump can be predicted using the predicted electric energy value, the predicted electric energy value can be predicted with higher accuracy.

Embodiment 3 FIG.
FIG. 19 is a block diagram showing the configuration of the power demand prediction apparatus according to Embodiment 3 of the present invention.
FIG. 20 is a diagram for explaining a neural network prediction model for obtaining a first power amount prediction value in the demand prediction unit of the power demand prediction apparatus shown in FIG.
FIG. 21 is a diagram for explaining a neural network prediction model for obtaining a second power amount prediction value in the demand prediction unit of the power demand prediction apparatus shown in FIG.
FIG. 22 is a flowchart for explaining the operation of the demand prediction unit of the power demand prediction apparatus shown in FIG.
The same parts as those in the above embodiments are given the same reference numerals, and the description thereof is omitted.

In FIG. 19, the power demand prediction apparatus 1 includes a first power demand record value DB 17, a pump maximum power amount value DB 18, a second power demand record value DB 19, and an air temperature value DB 20.
And the performance data acquisition part 50 stores the 1st electric power demand performance value EPN for every time of only the pump of the plant | facility 2 in 1st electric power demand performance value DB17.
Furthermore, the performance data acquisition unit 50 stores the second power demand performance value EEN for each time other than the pump of the facility 2 in the second power demand performance value DB 19.
Here, for each time, data for every hour is acquired as in the above embodiments.

Furthermore, the performance data acquisition unit 50 performs the first power demand performance at each time from the first power demand performance value EPN in the first power demand performance value DB 17 to the target time TN from the fifth predetermined time M5 before the target time TN. Among the values EPN, the maximum first power demand record value EPN is detected and extracted as the pump maximum power amount value GPN at the target time TN.
And the performance data acquisition part 50 is stored in pump maximum electric energy value DB18 as the pump maximum electric energy value GPN of the object time TN.

The weather information acquisition unit 30 acquires the weather information 3 including the actual temperature value and the predicted temperature value from the outside, and stores the temperature value UN for each time in the temperature value DB 20.
In addition, every time here is every time that is the same as the actual power demand values EPN and EEN, and data for every hour is acquired.
The precipitation of the weather information acquisition unit 30 is acquired in the same manner as in the first embodiment, and the past precipitation RN and the future precipitation QN at the target time TN are the same as in the first embodiment. Therefore, the description thereof is omitted as appropriate.

Next, operation | movement of the electric power demand prediction apparatus 1 of Embodiment 3 comprised as mentioned above is demonstrated. In addition, the part of the operation | movement similar to said each embodiment is abbreviate | omitted suitably.
First, the actual data acquisition unit 50 acquires the actual power demand actual value for each time of each facility such as a pump, a blower, lighting, and air conditioning installed in the facility 2. The obtained data is divided into a first power demand record value EPN for only the pump and a second power demand record value EEN for other than the pump. The first power demand record value EPN is stored in the first power demand record value DB 17 and the second power demand record value EEN is stored in the second power demand record value DB 19.

  And the performance data acquisition part 50 is 5th predetermined time M5 before the target time TN from the 1st power demand performance value EPN of 1st power demand performance value DB17, for example here 4 hours (4H) before the target time TN. -The 1st electric power demand track record value EPN of each time from M5 to object time TN is each extracted. And among each 1st electric power demand track record value EPN, the largest 1st power demand track record value EPN is detected and extracted as the pump maximum electric energy value GPN of the object time TN.

And the performance data acquisition part 50 is stored in pump maximum electric energy value DB18 as the pump maximum electric energy value GPN of the object time TN. In this way, with each time as the target time TN, the pump maximum electric energy value GPN at each time (target time TN) is detected and stored in association with each target time TN.
Since this method can be performed in the same manner as the maximum power amount value GN shown in the first embodiment, detailed description thereof is omitted.

Next, the weather information acquisition part 30 acquires the weather information 3 containing the actual value of the temperature value for every hour, and the forecast value of the temperature value from the outside, and stores the temperature value UN for every hour in the temperature value DB 20. .
The precipitation of the weather information acquisition unit 30 is acquired in the same manner as in the first embodiment, and the past precipitation RN and the future precipitation QN at the target time TN are the same as in the first embodiment. Therefore, the description thereof is omitted as appropriate.

Next, the demand prediction unit 40 will be described.
In the third embodiment, the power demand prediction is predicted separately for the pump of the facility 2 and the pumps other than the pump of the facility 2.
Using the values of the past precipitation DB 14, the future precipitation DB 15, the pump maximum power value DB 18, and the first power demand record value DB 17 for each past target time TN obtained as described above, the target time Build a prediction model for each TN.

  Then, the first electric energy prediction value FPX at the prediction target time TX is calculated from the past precipitation RX at the prediction target time TX, the future precipitation QX, and the pump maximum electric energy value GPX. Furthermore, the prediction model for every object time TN is constructed | assembled using each value of 2nd electric power demand performance value DB19 and temperature value DB20. Then, the second electric energy prediction value FFX at the prediction target time TX is calculated from the temperature value UX at the prediction target time TX. Then, the first power amount prediction value FPX and the second power amount prediction value FFX are summed to predict the power amount prediction value FX at the prediction target time TX.

  In addition, as a calculation method of the electric energy prediction value FX, the method of calculating using the prediction model by a multiple regression analysis or a neural network similarly to said each embodiment can be considered. A method using a pattern matching method is also conceivable.

First, as a specific example, a method for calculating the pump first electric energy prediction value FPX using a prediction model based on multiple regression analysis will be described.
A formula for calculating the first predicted electric power value FPX of the pump at the prediction target time TX using the prediction model is shown below as (Formula 9).

  First electric energy prediction value (FPX) = PP · Past precipitation (RN) + PF · Future precipitation (QN) + MAX · Pump maximum electric energy value (GPN) + intercept (Equation 9)

  Here, PP, PF, and MAX are partial regression coefficients, and intercept represents an intercept. The constant regression coefficient and the intercept are obtained by using a least square method so that a constant term that minimizes the sum of squares of errors in (Equation 10) shown below is obtained.

Next, a method of calculating the second electric energy prediction value FFX other than the pump using the prediction model based on multiple regression analysis will be described.
An equation for calculating the second electric energy prediction value FFX other than the pump at the prediction target time TX using the prediction model is shown below as (Equation 11).

Second electric energy prediction value (FFX) = temp · temperature value (U) + intercept
... (Formula 11)

  Here, temp is a regression coefficient, and intercept represents an intercept. For the regression coefficient and the intercept, a constant term that minimizes the sum of squares of errors in (Equation 12) shown below is obtained using the least square method.

  And the 1st electric energy prediction value FPX and the 2nd electric energy prediction value FFX which were calculated | required in this way are totaled, and the electric energy prediction value FX of the prediction object time TX of the facility 2 is calculated.

Next, as a specific example, a method for calculating a predicted electric energy value using a prediction model based on a neural network will be described.
FIG. 20 shows an example of a neural network prediction model for calculating the pump first electric energy prediction value FPX.

  As shown in FIG. 20, the first precipitation amount predicted value FPX at the prediction target time TX using the prediction model with the past precipitation RX at the prediction target time TX, the future precipitation QX, and the maximum pump power value GPN as input values. Is calculated. A formula for calculating the first electric energy prediction value FPX is shown below as (Formula 13).

  Here, Xj represents the input value, the past precipitation RX, the future precipitation QX, and the pump maximum power value GPN, wi is a coupling coefficient of each unit connecting the output layer to the intermediate layer, and wij is an intermediate value between the input layer and the output layer. The coupling coefficient of each unit connecting the layers, f is an output function in the output layer, and fj is an output function in the intermediate layer. Note that the coupling coefficients wi and wij of each unit are calculated by the method of least squares from the relationship between the past precipitation RN, the future precipitation QN, the maximum startup number VN, and the first power demand actual value EPN at each past time. To do.

FIG. 21 shows an example of a neural network prediction model for calculating the second electric energy prediction value FFX other than the pump.
As shown in FIG. 21, the second electric energy predicted value FFX at the prediction target time TX using the prediction model using the temperature value at the prediction target time TX, the highest temperature value on the same day, and the lowest temperature value on the same day as input values. Is calculated. A formula for calculating the second power amount prediction value FFX is shown below as (Formula 14).

  Here, Xj represents an input value, that is, an air temperature, a maximum temperature on the same day, and a minimum temperature on the same day, wi is a coupling coefficient of each unit connecting the intermediate layer to the output layer, and wii connects the input layer and the intermediate layer. The coupling coefficient of each unit, f represents an output function in the output layer, and fj represents an output function in the intermediate layer. The coupling coefficients wi and wij of each unit are calculated by the least square method from the relationship between the temperature at each past time, the highest temperature on the same day, the lowest temperature on the same day, and the second actual power demand value EEN.

  Then, the first power amount predicted value FPX and the second power amount predicted value FFX obtained in this way are summed to calculate the power amount predicted value FN at the prediction target time TX of the facility 2.

  Next, as a specific example, a method for calculating a predicted electric energy using a pattern matching method will be described. FIG. 22 is a flowchart showing a processing procedure of power demand prediction in the demand prediction unit 40.

First, a narrowing condition for acquiring the second power demand actual value EEN used for power demand prediction from the past second power demand actual value DB 19 is acquired (step ST21 in FIG. 22).
For example, as the narrowing-down condition, the following example using the temperature value UN can be considered.

| Temperature value UX (predicted value) at prediction target time TX-Air temperature value UN (actual value) at comparison target time TN | ≦ A7
The above formula is used as a narrowing condition.
A7 takes a value of 0 or more and differs depending on the facility configuration of facility 2 and the possible amount of wastewater treatment. Therefore, accuracy evaluation is performed using past data when the system is introduced, and the best value is set.

Next, based on the narrowing condition, the second power demand actual value EEN at the comparison target time TN corresponding to the narrowing condition is acquired from the second power demand actual value DB 19 (step ST22 in FIG. 22).
Next, an average value of the acquired second power demand actual value EEN is calculated, and this value is set as the second power amount predicted value FFX at the prediction target time TX (step ST23 in FIG. 22).

Next, a narrowing condition for acquiring the first power demand actual value EPN used for power demand prediction from the past first power demand actual value DB 17 is acquired (step ST24 in FIG. 22).
For example, the narrowing-down condition is performed in the same manner as in the first embodiment. Therefore, the description is omitted as appropriate. However, in the third embodiment, the pump maximum power amount value GPN is used instead of the maximum power amount value GN.

Next, based on the narrowing condition, the first power demand actual value EPN at the comparison target time TN corresponding to the narrowing condition is acquired from the first power demand actual value DB 17 (step ST25 in FIG. 22).
Next, an average value of the acquired first power demand actual value EPN is calculated, and this value is set as the first power amount predicted value FPX at the prediction target time TX (step ST26 in FIG. 22).
Next, the second power amount predicted value FFX calculated in the previous step ST23 and the first power amount predicted value FPX are summed to calculate a power amount predicted value FX at the prediction target time TX of the facility 2 ( Step ST27 in FIG.

  As described above, the first power amount predicted value FPX and the second power amount predicted value FFX at the prediction target time TX predicted by the demand prediction unit 40 using each method are the first prediction target time TX at the first target time TX. The power demand record value EPX is stored in the first power demand record value DB 17 and the second power demand record value EEX at the prediction target time TX is stored in the second power demand record value DB 19. The actual power demand values EPX and EEX stored based on the prediction of the prediction target time TX are the actual power demand actual values EPX and EEX obtained when the actual time reaches the prediction target time TX. When it can be obtained, the actual power demand actual values EPX and EEX are replaced and stored.

  According to the power demand prediction apparatus of the third embodiment configured as described above, the effects similar to those of the above-described embodiments can be obtained, as well as being greatly affected by precipitation and the operating state of the pump. By accurately calculating the predicted electric power amount of the pump that is expected to change and the predicted electric energy amount of the pump other than the pump that is expected to change due to the influence of temperature, the power demand that takes into account the rainwater of the facility is accurately predicted. can do.

  It should be noted that the present invention can be freely combined with each other within the scope of the invention, and each embodiment can be appropriately modified or omitted.

1 power demand forecasting device, 2 facilities, 3 weather information, 11 power demand record DB,
12 Maximum power value DB, 13 Precipitation DB, 14 Past precipitation DB,
15 Future precipitation DB, 16 Pump operation state DB, 17 First power demand record DB,
18 pump maximum electric energy value DB, 19 second external power demand actual value DB, 20 temperature value DB, 30 weather information acquisition unit, 40 demand prediction unit, 50 actual data acquisition unit,
60 Pump data acquisition unit, GN maximum energy value (at target time TN),
GY maximum electric energy value (at the prediction start time TY),
GX maximum energy value (at prediction target time TX),
EN Actual power demand (at target time TN),
EX Actual power demand value (at forecast target time TX),
EY Actual power demand value (at forecast start time TY),
FY Electric energy prediction value (at prediction start time TY),
FX power amount prediction value (at prediction target time TX), TN target time, T1 time,
TY prediction start time, TX prediction target time, M1 first predetermined time, M2 second predetermined time, M3 third predetermined time, M4 fourth predetermined time,
RN Past precipitation (at target time TN),
RY Past precipitation (at forecast start time TY),
RX Past precipitation (at forecast target time TX),
QN Future precipitation (at target time TN),
QY Future precipitation (at forecast start time TY),
QX Future precipitation (at forecast time TX), SN precipitation value, UN temperature value,
UX temperature value (at forecast target time TX), VN maximum number of starts, VY maximum number of starts,
VX maximum activation number, αL weight, βL weight, γL weight, αK weight, βK weight,
γK weight.

Claims (11)

In the power demand forecasting device that predicts the predicted amount of power at the forecast target time in a facility that performs sewage and rainwater treatment,
The actual power demand value at each time of the facility is acquired, the maximum actual power demand value from the first predetermined time before the target time to the target time is detected with respect to the actual power demand value, and the target time An actual data acquisition unit that extracts the maximum electric energy value of
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. The cumulative precipitation of each precipitation value at each time from the target time to the third predetermined time after the target time is calculated by calculating the cumulative precipitation of the quantity value and extracting it as the past precipitation at the target time. A meteorological information acquisition unit that calculates and extracts as future precipitation at the target time;
A maximum power amount value database for storing the maximum power amount value at the target time;
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
A power demand prediction device comprising: a demand prediction unit that predicts the predicted power amount at the prediction target time from the relationship between the maximum power amount value, the past precipitation amount, and the future precipitation amount, and the actual power demand value. In the power demand prediction device that predicts the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
A record data acquisition unit that acquires a power demand record value at each time of the facility;
A pump data acquisition unit that extracts the operation state of the pump from the fourth predetermined time before the target time to the target time as the target operation state;
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. A cumulative amount of precipitation is calculated and extracted as past precipitation at the target time, and the cumulative precipitation amount at each time from the target time to the third predetermined time after the target time is calculated. A meteorological information acquisition unit for extracting as future precipitation at the target time;
A pump operating state database for accumulating the target operating state at the target time;
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
A power demand prediction apparatus comprising: a demand prediction unit that predicts the predicted power amount at the prediction target time from the relationship between the target operating state, the past precipitation amount, and the future precipitation amount and the actual power demand value. 3. The pump data acquisition unit according to claim 2, wherein the target operation state is extracted as a maximum number of start of the pump or a total number of start of the pump from a fourth predetermined time before the target time to the target time. Electric power demand forecasting device. The said pump data acquisition part predicts the driving | running state of the said pump of the said prediction target time from the said electric energy prediction value of the said prediction target time predicted in the said demand prediction part. Power demand forecasting device. The said demand prediction part sets the said electric energy prediction value of the said prediction object time to the said performance data acquisition part as the said electric power demand actual value of the said prediction object time in any one of Claims 1-4. The power demand prediction apparatus described. In the power demand prediction device that predicts the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
Obtain the first power demand actual value of the pump and the second power demand actual value other than the pump of the facility,
A record of detecting the maximum first power demand record value from the fifth predetermined time before the target time to the target time with respect to the first power demand record value and extracting it as the pump maximum power amount value at the target time A data acquisition unit;
Obtaining weather information including precipitation values and temperature values of actual values and forecast values related to the facility, extracting a temperature value of the target time for the weather information, and a first value of the target time (2) calculating a cumulative precipitation of each precipitation value at each time from a predetermined time before to the target time and extracting it as a past precipitation at the target time; and a third predetermined time from the target time to the target time A meteorological information acquisition unit that calculates a cumulative precipitation of each precipitation value at each time until a time later and extracts it as a future precipitation at the target time;
A pump maximum electric energy value database for storing the pump maximum electric energy value at the target time;
A temperature value database for accumulating temperature values at the target time; and
A past precipitation database for accumulating past precipitation at the target time;
A future precipitation database for accumulating future precipitation at the target time;
Predicting the first electric power prediction value of the pump at the prediction target time from the relationship between the pump maximum electric energy value, the past precipitation, and the future precipitation, and the first electric power demand actual value, and the temperature value And a predicted value of the second power amount other than the pump at the prediction target time from the relationship between the second power demand actual value and the total value of the first power amount predicted value and the second power amount predicted value power demand prediction apparatus and a demand prediction unit configured to predict as the power amount prediction value. The demand prediction unit uses the first power amount predicted value at the prediction target time as the first power demand actual value at the prediction target time, and the second power amount predicted value at the prediction target time. The power demand prediction apparatus according to claim 6, wherein the second power demand actual value at the prediction target time is set in the actual data acquisition unit. The weather information acquisition unit
The past precipitation is calculated by weighting the precipitation value at each time from the second predetermined time before the target time to the target time,
The electric power according to any one of claims 1 to 7, wherein the future precipitation is calculated by weighting the precipitation value at each time from the target time to a third predetermined time after the target time. Demand forecasting device. In the power demand prediction method for predicting the predicted amount of electric power at the prediction target time in a facility that performs sewage and rainwater treatment,
The actual power demand value at each time of the facility is acquired, the maximum actual power demand value from the first predetermined time before the target time to the target time is detected with respect to the actual power demand value, and the target time Extracted as the maximum power value of
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. The cumulative precipitation of each precipitation value at each time from the target time to the third predetermined time after the target time is calculated by calculating the cumulative precipitation of the quantity value and extracting it as the past precipitation at the target time. Is calculated and extracted as future precipitation at the target time,
A power demand prediction method for predicting the predicted power amount at the prediction target time from the relationship between the maximum power amount value, the past precipitation amount, and the future precipitation amount and the actual power demand value. In the power demand prediction method for predicting the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
Obtain the power demand actual value at each time of the facility,
The operation state of the pump from the fourth predetermined time before the target time to the target time is extracted as the target operation state,
The weather information including the precipitation value of the actual value and the forecast value related to the facility is acquired, and each precipitation at each time from the second predetermined time before the target time to the target time is obtained with respect to the weather information. A cumulative amount of precipitation is calculated and extracted as past precipitation at the target time, and the cumulative precipitation amount at each time from the target time to the third predetermined time after the target time is calculated. And extract it as future precipitation at the target time,
A power demand prediction method for predicting the predicted power amount at the prediction target time from the relationship between the target operating state, the past precipitation amount, and the future precipitation amount and the actual power demand value. In the power demand prediction method for predicting the predicted amount of power at the prediction target time in a facility that has a pump and performs sewage and rainwater treatment,
The first power demand record value of the pump and the second power demand record value other than the pump of the facility are acquired, and the target from the fifth predetermined time before the target time with respect to the first power demand record value The maximum first power demand actual value until the time is detected and extracted as the pump maximum power amount value at the target time,
Obtaining weather information including precipitation values and temperature values of actual values and forecast values related to the facility, extracting a temperature value of the target time for the weather information, and a first value of the target time (2) calculating a cumulative precipitation of each precipitation value at each time from a predetermined time before to the target time and extracting it as a past precipitation at the target time; and a third predetermined time from the target time to the target time Calculate the cumulative precipitation of each precipitation value at each time until time later and extract it as future precipitation at the target time,
Predicting the first electric power prediction value of the pump at the prediction target time from the relationship between the pump maximum electric energy value, the past precipitation, and the future precipitation, and the first electric power demand actual value, and the temperature value And a predicted value of the second power amount other than the pump at the prediction target time from the relationship between the second power demand actual value and the total value of the first power amount predicted value and the second power amount predicted value power demand prediction method for predicting as the power amount prediction value.

Images